Patel et al (1) recently evaluated neuroradiologic features in a subset of their series of 78 spinal cord gangliogliomas, a series that triples the published cases extant only a few years ago (2). As their key premise, the authors claim they now view certain patterns of synaptophysin immunostaining as unique for diagnosis of gangliogliomas. They state that this synatophysin staining is always invoked when an astrocytoma vs. ganglioglioma diagnosis is disputed after thorough review of all available hematoxylin-eosin stains. They emphasize that in a 1993 study, “21 [gangliogliomas] originally diagnosed as astrocytomas at our institution were only recognized as gangliogliomas by using immunohistochemical analysis, and 23 of 25 gangliogliomas originally examined at outside institutions were called astrocytomas until we performed immunohistochemical studies.” Biopsies that are determined to be nondiagnostic for ganglioglioma vs. entrapped neurons, even in the eyes of experienced neuropathologists, are exactly the cases in which any interpretation contributed by positive neuronal synaptophysin staining is most likely to be deceptive (3, 4).
The authors fail to cite a detailed four-page study published 2 years ago that challenged the concept that any pattern of synaptophysin immunostaining was unique to spinal cord gangliogliomas (3). Instead, the authors refer to an incorrect edition of a general pathology textbook (their reference 51). They provide the reader with other potential cause to be skeptical of their synaptophysin criteria; an article that does not use synaptophysin staining is cited (their reference 53). To prove their contention that synaptophysin-positive neurons never occur in spinal cord white matter, and must always appear in a background of fine neuropil, the authors cite an abstract (5). The citations in this part of the paper are incorrect or irrelevant. Readers interested in the validity of synaptophysin staining for ganglioglioma diagnosis should refer directly to Zhang and Rosenblum; a more recent article of mine extends this study by analyzing synaptophysin immunostaining in the diencephalon and brain stem (4).
Patel et al argue that synaptophysin-positive neurons are unique to gangliogliomas. This is simply not the case. Synaptophysin-positive neurons are widely distributed in the normal spinal cord (3), and clusters of synaptophysin-positive neurons can be found literally embedded in white matter in the normal medulla and cervicomedullary region (4). Figure 1A–B shows single and clustered synaptophysin-positive neurons in the white matter of normal cervical spinal cord. Such neurons were easily found in adult cord obtained 6 hours postmortem; tissues were fixed for 4–6 hours to simulate fixation conditions likely to be encountered in the type of neurosurgical biopsy tissues immunostained by Patel et al. There is no question that native neurons, once entrapped in tumors, can be intensely synaptophysin-positive. Figure 1C–D illustrates the dorsal medulla of an adult woman with malignant lymphoma. This case shows large, synaptophysin-positive native neurons becoming entrapped as tumor infiltrates and effaces the medial nucleus ambiguous. Again, there is no synaptophysin-positive neuropil background as Patel et al claim must exist when native neurons are entrapped in infiltrating tumor (Fig. 1D). Thus, synaptophysin-positive neurons may be found in spinal cord gray matter (3), in white matter, and in unpredictably synaptophysin-negative or synaptophysin-positive background as various zones of spinal cord are infiltrated or effaced by gliomas.
The authors find that spinal cord gangliogliomas are the second-most common intramedullary spinal cord tumor. The authors propose that most spinal cord gangliogliomas show only limited neuronal foci and have previously been misdiagnosed as astrocytomas because of undersampling. The basic proposition seems unlikely, but other authors have found that only 15% of gangliogliomas have purely astrocytic zones (6). Large autopsy series of brain and spinal cord tumors have never reported higher rates of gangliogliomas (7). Some of the neuroradiologic findings of Patel et al can be explained. The most biologically infiltrative astrocytomas, studied in increasingly large resections, are more likely to engulf native neurons focally, a circumstance that is congruent with the authors' findings of only focal neoplastic neurons in their numerous gangliogliomas. Tumors less likely to show edema on neuroimaging would possibly be more likely to maintain surviving, entrapped, synaptophysin-positive native neurons at least focally within a large resection volume. Thus, a ganglioglioma diagnosis could select for the larger and more freely infiltrative tumors that might share characteristic neuroimaging and high-recurrence patterns. Patel et al mention that, in their series of 174 spinal cord tumors, not one pilocytic astrocytoma was recognized. Nonetheless, the most circumscribed, least infiltrative astrocytomas (perhaps those entities interpreted by other groups as common spinal cord pilocytic astrocytomas) would be less likely to show entrapped neurons, and would have different imaging and recurrence characteristics.
- Copyright © American Society of Neuroradiology